ERIC EJ965801: Undergraduate Borrowing and Its Effects on Plans to Attend Graduate School Prior to and after the 1992 Higher Education Act Amendments

2006·0.11 MB·English

by  ERIC

Undergraduate Borrowing and Its Effects on
Plans to Attend Graduate School
Prior to and After the 1992 Higher Education
Act Amendments
By Dongbin Kim and Therese S. Eyermann
Dongbin Kim is an assistant As student loan indebtedness has more than doubled in the past
professor in the Department decade, it has become important to examine the effects of under-
of Teaching and Leadership
graduate debt on graduate school attendance. The significant
in the School of Education at
the University of Kansas. increase in student borrowing can be attributed primarily to the
Therese S. Eyermann is passage of the Higher Education Amendments of 1992, which
executive assistant to the increased federal student loan limits and expanded eligibility in
president for California State
student loan programs. To measure the effects of the increased
University, Channel Islands
undergraduate borrowing on graduate school attendance, this
study compared graduate school plans of students who attended
colleges and universities between 1985-1989 (before the Amend-
ments) and 1994-1998 (after the Amendments). The results indi-
cated a slight negative effect of borrowing on students’ plans to
attend graduate schools prior to the 1992 Amendments. By con-
trast, the students who attended school after the 1992 Amend-
ments showed significant positive effects of borrowing, particu-
larly for middle-income students. If middle-income students bor-
rowed money, they were more likely to plan to attend graduate
school than students from high-income families with loans.
I
n 2005, nearly 69% of high school graduates were enrolled
in colleges and universities in the following fall semester (U.S.
Bureau of Labor Statistics, 2006). As the college enrollment
rate for high school students has reached a historical high since
1959, some wonder if the trends in undergraduate enrollment
will translate to the graduate arena. Therefore, just as the pipe-
line of effects that lead a person to attend college have been
studied, it is important to examine the factors that may affect
graduate school attendance. Specifically, what are the effects of
undergraduate student loans on a student’s attendance in gradu-
ate school and have those effects changed over time?
Student loans, initially only 18% of total financial aid in
1976-77, grew to 58% of all aid in 1996-97. In 2003-04, loans
constituted 56% of aid, remaining the most common type of aid
(The Education Resources Institute, 1998; The College Board,
2004). This explosion in student borrowing can be attributed
primarily to the passage of the Higher Education Amendments
of 1992, which created the Federal Unsubsidized Stafford Loan
and Federal Direct Unsubsidized Loan programs and increased
loan limits to meet the continual rise in the cost of attending
higher education. By expanding eligibility, the unsubsidized
NASFAA JOURNAL OF STUDENT FINANCIAL AID 5
Stafford and Direct Loans dramatically increased the amounts
students could borrow, as well as the number of borrowers,
including students from middle- and upper-income families (The
Education Resources Institute, 1998). For example, of full-time,
full-year dependent undergraduates, 30% had borrowed in 1989-
90 but 45% had borrowed in 1999-2000 (National Center for
Education Statistics, 2005). The average loan amount borrowed
per full-time equivalent (FTE) increased by 200%, from $1,957
in 1989-90 to $5,840 in 1999-2000, in constant 2003 dollars
(The College Board, 2004). As the number of borrowers and the
amounts borrowed have increased, it is pertinent to examine
how this undergraduate borrowing may have affected graduate
school attendance. At this point, however, only a very small body
of literature addresses the relationship between financial aid
and graduate school attendance (Heller, 2001). Therefore, it
seems appropriate to examine the effects of loan debt on gradu-
ate school attendance, particularly as borrowing has become a
primary financing tool for today’s college students (NCES, 2005;
Heller, 2001; Baum & O’Malley, 2003).
Literature Review In higher education, the role of financial aid has been framed in
terms of access to higher education (i.e., whether to enroll in
college) and choice (i.e., which type of college to attend). How-
ever, theorists such as Rawls (1971) and Le Grand (1982) have
argued that equity cannot be realized until one examines the
final outcomes. Eyermann (1999) supported their argument,
saying that equity can only be achieved when students who have
similar ability and motivation attain comparable outcomes, re-
gardless of their financial situations. Therefore, assessing equity
demands that one looks at longer-term outcomes, including
graduate school attendance.
As financial aid policy has moved rapidly toward a loan-
based system, it is important that higher education researchers
examine the ramifications of this method of funding from an
equity viewpoint. For example, if loans encourage graduate
school enrollment, then they could be serving a valuable pur-
pose in meeting the goal of equal access. However, if loans
serve as a deterrent to graduate enrollment, particularly for
at-risk or under-represented students, then the current loan-
based financial aid policy needs to be reexamined.
In general, studies on graduate school participation have
not focused on financial aid. Of the few that have, findings indi-
cate that student background and collegiate factors other than
aid play a more significant role. These same findings indi-
cate that there is little association between the amount of debt
and the decision of students to continue their education (Heller,
2001; Schapiro, O’Malley, & Litten, 1991; Weiler, 1994; Millett,
1999). Focusing on students who received a bachelor’s degree in
the 1992-93 academic year, Heller (2001) examined the rela-
tionship between undergraduate indebtedness and a students’
6 VOL. 36, NO. 2, 2006
decisions to attend graduate school. In contrast to the signifi-
cant effects of degree aspirations, major, and college GPA on
graduate school enrollment, this study found little impact of
undergraduate borrowing. Although the insignificant relation-
ship between loans and graduate school enrollment (i.e., loans
do not necessarily hamper a student’s decision to attend gradu-
ate school) supports the current loan-based financial aid policy,
Heller concluded that the student sample in the study had at-
tended college before the Higher Education Amendments of 1992,
and thus the study may not reveal the direct impact of the cur-
rent loan-based aid policy on graduate school enrollment.
Schapiro et al. (1991), in their examination of the rela-
tionship between debt and plans to attend graduate school in
the fields of arts and sciences, found that while gender and
ethnicity were significantly related to graduate school attendance,
debt was not. However, their data, collected by the Consortium
on Financing of Higher Education (COFHE) Senior Survey, were
exclusively drawn from students at elite colleges and universi-
ties who are members of the COFHE. COFHE is an institution-
ally supported organization of thirty-one private colleges and
universities including Brown University, Columbia University,
Harvard University, Michigan Institute of Technology, Princeton
University, Stanford University, and others. Higher proportions
of upper-income students with higher academic ability attend
these elite institutions. If, as previous research suggests, stu-
dent college outcomes are mitigated by socioeconomic status
(SES) level and high school academic experiences, then this
choice of data set reduces the opportunity to determine the ef-
fects of loans on lower SES levels.
Using 1980 High School and Beyond data from the Na-
tional Center for Education Statistics (NCES), Weiler (1994) found
that undergraduate debt had no association with the decision
of a college graduate to attend graduate school. This study con-
trolled for SES background characteristics as well as degree
aspirations, ability, and major. Although seniors who planned
to attend graduate school had smaller debts than those who
were only working toward a baccalaureate degree, the multi-
variate analysis did not find any significant effect of debt on the
decision to attend graduate school.
Weiler (1994) noted that debt levels may relate more to
intentions than to actual enrollment in graduate school. He
speculated that future plans play a minimal role at the times
the student borrows as an undergraduate, and only gradually
come into play as the student considers options of successively
higher levels of debt required by graduate school. This study,
however, also focused on the students who borrowed between
1980 and 1984, the early stages of the shift from grants to loans.
Therefore, the effects of loans from this study do not necessarily
reflect the possible differences after the Higher Education Amend-
ments of 1992.
NASFAA JOURNAL OF STUDENT FINANCIAL AID 7
Millett (1999) performed a logistic regression analysis
using Baccalaureate and Beyond data from NCES to determine
the odds of attending graduate school. Debt was not found to be
a significant factor. However, the study failed to control for stu-
dents’ early aspirations and motivations as well as their career
intentions.
Research This study investigates the plans to attend graduate school from
Questions two separate college cohorts. The first cohort attended colleges
and universities between 1985 and 1989, prior to the expan-
sion of loans. The second cohort attended colleges and universi-
ties between 1994 and 1998, after the passage of the Higher
Education Amendments of 1992. Research questions raised in
the study are as follows:
1. Is the total amount of student loans borrowed as an
undergraduate related to students’ plans to attend gradu-
ate school?
2. Is there any significant difference in the factors that in-
fluence students’ plans to attend graduate school before
and after the 1992 Higher Education Amendments, with
a particular focus on the effect of the total amount of
loans?
Methodology Data Source
This study used longitudinal data collected by the Cooperative
Institutional Research Program (CIRP) of the Higher Education
Research Institute (HERI) at the University of California at Los
Angeles. The student survey data sets were collected at two time
points: at the time of college entry and four years later. Stu-
dents at all institutions of higher education listed in the Fall
Enrollment Files of the U.S. Department of Education’s Inte-
grated Postsecondary Education Data System (IPEDS) were the
targeted population for the CIRP survey. Given that the compre-
hensive longitudinal database offers a unique set of student and
family background characteristics, college environmental vari-
ables, and post-collegiate outcomes, the data allow comprehen-
sive examination of the effects of college loans in this study.
Two data sets were used for this study. The first sample
consisted of students who were surveyed originally as freshmen
in 1985 and again in 1989. The second sample consisted of
students who were surveyed as freshman in 1994 and resur-
veyed in 1998. The follow-up surveys of each group, conducted
four years after college entry, provided information about stu-
dents’ plans to attend graduate school and the amount of loans
borrowed during their undergraduate years.
In 1985, 280,000 college students from 546 institutions
participated in the CIRP Freshman survey (Astin et al., 1985).
Of those, a randomly selected subsample of 86,000 students
was requested for the follow-up survey in 1989. With a response
8 VOL. 36, NO. 2, 2006
rate of 28.8%, the follow-up survey resulted in a total response
of 24,847 students from 309 institutions (Astin, 1993).
The 1994 CIRP freshmen survey included 237,777 first-
year students from 461 colleges and universities. Using a simi-
lar stratified random sampling, the 1998 follow-up survey pro-
duced a total response of 16,078 students from 154 colleges
and universities. In both 1989 and 1998, the surveys were con-
Whether students ducted for randomly-selected students who had responded to
start their the freshman survey, regardless of the students’ college enroll-
postsecondary ment status (e.g., whether they enrolled in the same school in
education in four- their fourth year as their first enrolled or whether or not the
students left the institutions temporarily or permanently).
year or two-year
Participation bias relating to CIRP sampling techniques
institutions
included institutional selectivity, type, and control, as well as
influences students’
students’ individual characteristics including race and gender.
college experiences
Participation bias was adjusted using the CIRP stratification
and various college
scheme (Astin et al., 1994).
outcome measures, Whether students start their postsecondary education
including persistence in four-year or two-year institutions influences students’ col-
rates, bachelor’s lege experiences and various college outcome measures, includ-
degree completion ing persistence rates, bachelor’s degree completion rates, or
rates, or graduate graduate school enrollment rates. For example, only about 20%
of the students who first attended public 2-year institutions at-
school enrollment
tained bachelor’s degrees or were still enrolled pursuing
rates.
bachelor’s degrees six years after college entry, in contrast to
the 70% of the students who started at 4-year institutions (NCES,
2003). Therefore, to limit any possible confounding factors that
influence the amount of loans borrowed and plans to attend
graduate school, this study focused exclusively on the students
who began their postsecondary education at four-year institu-
tions in 1985 and in 1994. Additionally, this study only exam-
ined traditional college-age students, students who did not de-
lay college entry, and students with no missing responses for
the predictors in both of the freshman and follow-up surveys.
Thus, the selected sample used for this study was 7,588
students for the first cohort and 5,234 students for the second
cohort. Among the first cohort group, 1,535 students (20%)
planned to attend graduate school and 6,053 (80%) did not. For
the second cohort, 1,114 students (21%) planned to attend
graduate school and 4,120 students (79%) did not.
Variables
The key independent variable was the amount of loan debt ac-
cumulated during the four years of undergraduate study, as
measured in 1989 and in 1998. To obtain an accurate estimate
of student undergraduate borrowing, the follow-up survey in
1998 asked students to report the total amount of student loans
they had borrowed for their undergraduate education. The 1989
survey, however, did not ask students to report their total amount
of undergraduate loans. To acquire an estimate for the amount
NASFAA JOURNAL OF STUDENT FINANCIAL AID 9
of loans in 1989, a new variable was computed based on stu-
dent responses in 1985 (see Appendix for details). The loan
amount measure was a continuous variable, coded by every
$1,000.
Individual characteristic variables included parental
income, student race/ethnicity, mother’s and father’s level of
education, high school academic achievements (as measured
by the high school grade point average), and degree aspirations
as a freshman. According to Kim (2003), student socio-economic
status (SES) has non-linear effects on student degree attain-
ment. The probability of degree attainment for low-SES students
To test the possible was 24.7%, which was significantly lower than 60% for high-
distinct effects of SES students. Although the negative tendency was less seri-
loans on graduate ous, the probability of middle-SES students completing a de-
school attendance gree was 38%, which was 22% lower than that of the reference
by parental income, students, who were high-SES.
the analysis Therefore, to examine the possible non-linear effects of
income on graduate school attendance, parental income was
included interaction
divided into three levels based on a frequency distribution: low-
terms between the
income ($0-$29,999), middle-income ($30,000-$59,999), and
amount of loans and
high-income ($60,000 or higher) for the 1985-89 cohort and
income.
low-income ($0-$29,999), middle-income ($30,000-$74,999) and
high-income ($75,000 or higher) for the 1994-98 cohort.
In addition, to test the possible distinct effects of loans
on graduate school attendance by parental income, the analy-
sis included interaction terms between the amount of loans and
income. Mothers’ and fathers’ education levels were used as
separate variables to examine any distinct effects of parental
education levels on students’ graduate school plans.
High school GPA and degree aspirations were used as
pre-college experience variables, which determine students’ col-
lege experiences and thus influence their plans to attend gradu-
ate school. College major and college GPA were included as col-
lege experience variables. College major is an important vari-
able because it often determines students’ exposure to gradu-
ate education and inclination to pursue further education. A
NCES study (1996) using Baccalaureate and Beyond Longitudi-
nal data indicates that college graduates who majored in the
arts and sciences were 9% more likely to enroll in further edu-
cation than those who majored in fields such as business, man-
agement, or education. The college GPA was dichotomous indi-
cator based on a four-point scale with grades of “B- or less,”
“B,” “B+ or A-,” being compared to the reference group of “A or
A+.”
Lastly, institutional characteristics, college tuition, and
undergraduate admission selectivity (as measured by the aver-
age SAT verbal and math scores of the entering freshmen for
the institution), were included in the statistical analysis.
10 VOL. 36, NO. 2, 2006
Statistical Analysis
Two logistic regression analyses were conducted to examine the
effect of loan indebtedness on the decision to attend graduate
school for the 1985-1989 cohort and the 1994-1998 cohort.
Logistic regression is an appropriate multivariate technique when
the outcome variable is dichotomous. In particular, when the
distribution of the outcome variable is highly skewed and can-
not satisfy the normality assumption of ordinary least square
regression, logistic regression is an appropriate technique. In
this study, about one-fifth of the students in both cohorts
planned to attend graduate school, while four-fifths of students
did not (Kleinbaum, Kupper & Muller, 1988). The B coefficients
for each variable in the logistic regressions were converted to
Delta-p statistics, using a mathematical equation recommended
by Peterson (1985):
L
e
1
Δp = L - p (p = “estimated population means” in Table 1)
1 + e
1
p
L = Ln + b (b = regression coefficient)
1 1 - p
p*(1 - 0)*e(b) - 1
Δp = 1 + p*e(b) - 1
The Delta-p statistics identify the relationship between
a unit change in a predictor and the estimated percentage change
in the outcome variable (Peterson, 1985; Cabrera, 1994). For
example, Delta-p statistics of .05 indicate that a unit change in
the predictor indicates a 5% increase in the probability of
students’ plans to attend graduate school.
Limitations One significant limitation of this study is that the follow-up
surveys were conducted four years after the students’ college
entry and they did not provide information on students’ actual
enrollment in graduate school but included their plans to at-
tend graduate school. Although not all students who plan to
attend graduate school actually enroll in graduate school, addi-
tional analysis presents a strong correlation between students’
plans to attend graduate school and their actual attendance.
The first cohort that entered colleges and universities in 1985
was resurveyed in 1994, 10 years after their first college entry.
This follow-up data offered a variable on students’ actual gradu-
ate school enrollment for the first cohort. The second cohort, for
1994-98, did not have a second follow-up survey, and thus it
was not possible to use the actual graduate school enrollment
for the study. Of the students who planned to attend graduate
school in 1989 (N = 1,503) however, 93.5% (1,406) actually at-
tended graduate school by 1994, indicating a strong correlation
of .423 at the .001 significance level.
NASFAA JOURNAL OF STUDENT FINANCIAL AID 11
A second limitation is that the loan amounts used in the
study were students’ self-reported measures. Students’ recol-
lections of the amounts borrowed for their undergraduate years
are not always accurate, thus the loan amount might be overes-
timated or underestimated by students’ tendencies in answer-
ing the survey. Additionally, the loan amount in the study in-
cluded all types of loans available through federal, state, insti-
tutional, and private loan programs, and parental borrowing
through the Parental Loans for Undergraduate Students (PLUS)
program. Although it is important to consider the type of loans
that students borrowed for their undergraduate education, be-
cause of the different payment plans and interest rates for each
type of loans, this study cannot show the impact of specific types
of loans on students’ plans to attend graduate school.
Results Results for the 1985-89 Cohort
The table on page 13 presents the Delta-p statistics for the pre-
dictors that have a statistically significant relationship with stu-
dents’ plans to attend graduate school. The second column in
the table presents the results of the logistic regression model
for the 1985-1989 cohort. In terms of model fit, 68% of all stu-
dents entered in the model were correctly classified. The good-
ness-of-fit statistics show that the model fits the data well, and
is statistically significant (df = 35, chi-square = 207.601) at the
p < .001 level.
The 1989 loan debt variable did not have a statistically
significant effect on students’ plans to attend graduate school
and the non-significant effect of loan debt was consistent across
students’ parental income. This finding confirms previous re-
search, which indicates that there is no significant (or marginal)
impact of loans on graduate school attendance for the students
who attended colleges and universities before the Higher Edu-
cation Amendments of 1992 (e.g., Heller, 2001; Baum &
Saunders, 1998; Weiler, 1994).
By contrast, various student background characteris-
tics and college experience variables had significant relation-
ships with students’ plans for graduate school attendance. Race/
ethnicity was a significant predictor of students’ plans to attend
graduate school: Asian students were 18% more likely to plan
to attend graduate school than their White, non-Hispanic coun-
terparts. Research has continuously shown that Asian students
tend to have higher college enrollment rates, persistence rates,
and graduate school attendance rates than other racial/ethnic
groups (e.g., NCES, 2005; Perna, 2004).
Parental income or father’s and mother’s education was
not a significant predictor of planning to attend graduate school.
However, students with a high school GPA of B+ or A- were sig-
nificantly less likely to plan to attend graduate school than their
counterparts with high school GPAs of A or A+. Students with
college GPAs of B- or lower and with a B average were 11% and
12 VOL. 36, NO. 2, 2006
Table
Delta-p Statistics of Borrowers Planning to Attend Graduate School,
by Cohort Groups
Variable 1985-89 cohort 1994-1998 cohort
Race (reference: White)
African American
Asian .18** .09**
Latino
Parental income (reference: High income)
Low Parental Income -.08*
Middle Parental Income -.09**
Father’s education (reference: Some graduate education
or graduate degree)
High school degree or less
Some college education or college degree
Mother’s education (reference: Some graduate education
or graduate degree)
High school degree or less
Some college education or college degree
High school GPA (reference: A or A+)
B or lower
B+ or A- -.05** -.03*
Degree aspirations as a freshman (reference: Professional
degrees such as MD or JD)
Bachelor degree -.13*** -.19***
Master or doctorate degree -.10*** -.12***
College GPA (reference: A or A+)
B- or less -.11*** -.16***
B -.06* -.12***
A- or B+ -.08***
College tuition (reference: High tuition)
Low tuition
Middle tuition
College selectivity (reference: High-top third)
Low (bottom third)
Middle (middle third) .05*
College major (reference: Other majors)
Biological science .20*** .10**
Business -.11***
Education -.10**
Engineering
English
Health professional .11* .09**
History/Political science .10*
Humanities
Fine arts
Math/Statistics
Physical science .25*** .16**
Social science
Amount of loans ($thousands)
Loan for high parental income
Loan for middle parental income .004*
Loan for low parental income
Estimated population means (percent planned to attend
graduate school) 20% 21%
Chi-square 207.601*** 477.528***
Percent of cases properly classified 68% 81%
Note. Only the variables that had statistically significant relationship with students’ plans to attend graduate school are presented.
* p < .05 level, ** p < .01 level, *** p < .001
NASFAA JOURNAL OF STUDENT FINANCIAL AID 13
6%, respectively, less likely to plan to attend graduate school
than their counterparts who had a college GPA of A or A+. There
was no significant difference in the probability of planning to
attend graduate school between students with GPAs of B+ or A-
and those with A or A+.
The degree aspirations as freshmen were another sig-
nificant predictor. Students who had low degree aspirations as
freshmen (bachelor’s degree) were 13% less likely to plan to at-
tend graduate school than students who had high degree aspi-
rations (professional degrees such as medicine or law). Students
with master’s or doctoral degree aspirations were 10% less likely
to plan to attend graduate school than their counterparts who
had professional degree aspirations.
Given that the Students’ major was also a statistically significant pre-
Higher Education dictor. Those who majored in biological sciences, the health pro-
Amendments of fessions, history/political sciences and physical sciences were
1992 introduced significantly more likely to plan to attend graduate school than
students whose major was in other subjects. In particular, stu-
unsubsidized loans,
dents whose majors were biological and physical sciences were
which have mainly
20% and 25%, respectively, more likely to attend graduate school
targeted middle-
than students of other majors. Students majoring in business
income families, the
or education were least likely to plan to attend graduate school.
small but significant
positive association
Results for the 1994-98 Cohort
between borrowing The third column in the table presents the results of the second
and planning to cohort logistic regression analysis. In terms of model fit, overall
attend graduate 81% of the students entered in the logistic regression model
school for middle- were correctly classified. The goodness-of-fit statistics show that
income under- the model fits the data well, and is also statistically significant
graduates clearly (df = 35, chi-square = 477.528) at the p < .001 level. The logistic
regression that predicts the likelihood of graduate school atten-
reflects the current
dance in 1998 revealed a different story from that of 1985-1989
emphasis in loan
cohort. In contrast to the non-significant effect of the 1989 loan
policy.
debt on graduate school attendance, the 1998 loan debt was a
significant positive predictor on graduate school attendance,
particularly for middle-income students. For the students from
low- and high- income groups, borrowing did not appear to be a
factor for their plans to attend a graduate school. However, for
middle-income students, borrowing had a slightly positive ef-
fect (0.4%) on plans to attend graduate school. This finding sup-
ports Eyermann’s (1999) argument that the profusion of college
loans after the 1992 Amendments might have created a culture
in which loans are now one of the most feasible ways to pay for
college. Given that the Higher Education Amendments of 1992
introduced unsubsidized loans, which have mainly targeted
middle-income families, the small but significant positive asso-
ciation between borrowing and planning to attend graduate
school for middle-income undergraduates clearly reflects the
current emphasis in loan policy.
14 VOL. 36, NO. 2, 2006